Study on a Filled Arithmetic Based on Random Sets for Data Mining of Aero-Craft Fault

Data mining has been applied to aircraft fault diagnosis. Data preprocessing is one of the key problems of data mining. Missing data brings difficulties to data preprocessing. In view of the randomness and uncertainty characteristics of missing aircraft fault data, the missing attribute parameter values were taken as random sets, a filled algorithm based on random sets was put forward from the standpoint of set theory. Through a data filled example of a real and overhauled civilian aero-engine, validity of the algorithm has been verified. The algorithm is adaptable to fill the fuzzy, continuous and missing data of aircraft fault parameters.